JP4860458B2 - Image evaluation apparatus and method, and program - Google Patents

Description

  The present invention relates to an image evaluation apparatus and method for evaluating an image using a face included in the image, and a program for causing a computer to execute the image evaluation method.

  In recent years, the spread of digital cameras and the increase in the capacity of recording media for storing images have made it possible for users to hold many images in one medium. . On the other hand, there is a trouble of selecting an image to be processed such as printing from a large number of images, and in order to select an image efficiently, the user finally prints by evaluating the image. There is a demand for a function of narrowing down an image under a certain condition before determining it as an image to be printed, and a function of selecting an image suitable for printing in accordance with user preferences.

  For example, Patent Document 1 discloses a method of evaluating an image based on any one of image brightness, an acceleration sensor, and AF evaluation, and automatically omitting a photograph inappropriate for printing using the evaluation result. Yes.

  In Patent Document 2, the orientation of a person's face included in an image is determined, an evaluation value of the image is calculated based on the determination result, and a desired image is selected from a plurality of images based on the calculated evaluation value A technique is disclosed.

Furthermore, Patent Document 3 discloses a method for evaluating an image using an evaluation result obtained by integrating a plurality of evaluation items such as a ratio of a face in an image, a degree of opening of an eye, a face direction, focus, blur, and brightness. It is disclosed.
JP 2002-10179 A JP 2004-361989 A JP 2002-358522 A

  As described above, various methods for evaluating an image based on a face included in the image have been proposed. However, since the methods described in Patent Documents 1 to 3 perform evaluation of an image by simply calculating an evaluation value for each evaluation item, subjective evaluation of a person who actually viewed the image is performed. It is not necessarily reflected, and the evaluation of the person who viewed the image may be different from the calculated evaluation.

  The present invention has been made in view of the above circumstances, and an object of the present invention is to evaluate an image with higher accuracy by using face information included in the image.

An image evaluation apparatus according to the present invention includes an information acquisition unit that acquires information on the size and position of a face from an image including at least one face;
Evaluation value calculation means for calculating an evaluation value that statistically represents the evaluation result of the image based on the face size and position information is provided.

  Specifically, the evaluation value calculation means refers to a first weighting function determined in advance according to the size and position of various faces, and the first value function according to information on the size and position of the face. What is necessary is just to use the 1st weighting coefficient calculated | required from the weighting function as said evaluation value.

  The “evaluation value” is not a value that can be quantitatively calculated from an image such as an S / N ratio or resolution, but a prediction calculated so as to correlate with a possible evaluation level of a user who wants to evaluate the image. Mean value.

  Here, “statistical” is preferably selected from a large number of sample images, rather than a priori based on several hypotheses when calculating a weighting function for calculating an evaluation value. This means that the image taken as the correct answer data is obtained inductively. The correct data may be selected in any way, but it is preferable to use correct data actually selected by the subject. The number of sample images for obtaining the weighting function is 300 or more, preferably 1000 or more.

In the image evaluation apparatus according to the present invention, the information acquisition means is means for further acquiring information on the number of faces.
The evaluation value calculation unit may be a unit that calculates the evaluation value based on information on the number of faces.

  In this case, the evaluation value calculation means refers to a second weighting function determined in advance according to the number of various faces, and from the second weighting function according to the information on the number of faces. And calculating a temporary evaluation value that is an evaluation value for the individual face by weighting the first weighting coefficient for each face with the second weighting coefficient. The temporary evaluation value calculated for the face may be added to calculate the evaluation value.

  Further, the evaluation value calculation means refers to a third weighting function determined in advance according to the size and number of various faces, and determines the first value according to the face number information and the face size information. A third weighting factor is obtained from the weighting function of 3, and the first weighting factor is weighted with respect to the individual face by the third weighting factor, thereby providing a temporary evaluation that is an evaluation value for the individual face. The evaluation value may be calculated by calculating a value and adding the temporary evaluation values calculated for all the faces.

Further, in the image evaluation apparatus according to the present invention, the information acquisition unit is a unit that acquires information on a distance from the face located at the center of the image of the face included in the image,
The evaluation value calculation unit may be a unit that calculates the evaluation value based on the distance information.

In this case, the evaluation value calculation means refers to a fourth weighting function determined in advance according to the distance from the face located at the center of the image, and determines the fourth weighting function according to the distance information. A fourth weighting factor is obtained from a weighting function, the first weighting factor or the temporary evaluation value is weighted according to the distance from the face closest to the center of the image, and the weighted first The weight coefficient may be calculated as the evaluation value, or the evaluation value may be calculated by adding the weighted temporary evaluation values calculated for all the faces.

In the image evaluation apparatus according to the present invention, the information acquisition unit is a unit that further acquires information on the face orientation,
The evaluation value calculation means may be a means for calculating the evaluation value based on the information on the face orientation.

  In this case, the evaluation value calculation means refers to a fifth weighting function determined in advance according to various face orientations, and from the fifth weighting function according to the information on the face orientation. 5 weighting factors, weighting the first weighting factor, the temporary evaluation value, or the first weighting factor weighted by the fourth weighting factor with the fifth weighting factor, and weighting Calculating the first weighting factor weighted by the first weighting factor or the fourth weighting factor as the evaluation value, or the weighted temporary evaluation calculated for all the faces The evaluation value may be calculated by adding the values.

An image evaluation method according to the present invention obtains information on the size and position of a face from an image including at least one face,
An evaluation value that statistically represents the evaluation result of the image is calculated based on the information on the size and position of the face.

  The image evaluation method according to the present invention may be provided as a program for causing a computer to execute the image evaluation method.

  According to the present invention, face size and position information is acquired from an image including at least one face, and an evaluation value that statistically represents an image evaluation result is calculated based on the face size and position information. . For this reason, it is possible to reflect the average preference of the person who views the image in the evaluation value, and thereby the image can be evaluated with higher accuracy.

  Further, if an image is selected from a plurality of images based on the evaluation value, the image can be selected more appropriately while reflecting the average preference of the viewer of the image.

  Further, by acquiring more information on the number of faces and calculating an evaluation value based on the information on the number of faces, the image can be evaluated with higher accuracy.

  Further, by further acquiring distance information from the face located at the center of the face included in the image and further calculating an evaluation value based on the distance information, the image can be evaluated with higher accuracy.

  Further, by further acquiring face orientation information and calculating an evaluation value based on the face orientation information, the image can be evaluated with higher accuracy.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a schematic block diagram showing the configuration of an image evaluation apparatus according to an embodiment of the present invention. As shown in FIG. 1, the image evaluation apparatus 1 according to the present embodiment performs various controls such as recording control and display control of image data, and controls each part of the apparatus 1 and operates the CPU 12. A system memory 14 composed of a program, viewer software for viewing images, a ROM in which various constants are recorded, and a RAM serving as a work area when the CPU 12 executes processing, and various instructions to the apparatus 1 An input unit 16 including a keyboard and a mouse for performing the display and a display unit 18 including a liquid crystal monitor for performing various displays.

  The image evaluation apparatus 1 also includes a card slot 20 for reading image data from the memory card 2 on which the image data is recorded and recording image data on the memory card 2, and a technique represented by JPEG for the image data. A compression / decompression unit 22 for decompressing or decompressing compressed image data, a hard disk 24 for recording image data, various programs executed by the CPU 12, a weighting table statistically obtained as described later, and a system memory 14, a memory control unit 26 that controls the card slot 20 and the hard disk 24, and a display control unit 28 that controls the display of the display unit 18.

  The image evaluation apparatus 1 also includes an information acquisition unit 30 that acquires information about a face from an image, and an evaluation value calculation unit 32 that calculates an evaluation value that represents an evaluation result of the image based on the information acquired by the information acquisition unit 30. With.

  Hereinafter, functions of the information acquisition unit 30 and the evaluation value calculation unit 32 will be described together with processing performed in the present embodiment. In the following description, it is assumed that the image is read from the memory card 2 and stored in the hard disk 24.

  FIG. 2 is a flowchart of the image evaluation process performed in the first embodiment of the present invention. Here, in the first embodiment, the evaluation value is calculated based on the face size and position information. When the user instructs to start image evaluation from the input unit 16, the CPU 12 starts processing. First, the image is read from the hard disk 24 (step ST1), and the initial value of the evaluation value is set to 0 (evaluation value = 0). Step ST2). Next, the information acquisition unit 30 detects a human face area from the image (step ST3). Specifically, for example, a rectangular region having a skin color in the image and surrounding the contour shape of the face is detected as the face region. In addition, you may detect the area | region which makes the outline shape of a face as a face area. Further, the method for detecting the face area is not limited to this. Here, when a plurality of faces are included, all face areas are detected.

  Next, the information acquisition unit 30 determines whether or not a face area is included in the image (step ST4), and if step ST4 is negative, the process ends. If step ST4 is affirmed, information on the size and position of the face included in the image is acquired from the detected face area (step ST5). As the face size, the number of pixels in the face area or the ratio of the face area to the entire image can be used.

  As the face size information, information representing the size of the face area in three levels of large, medium and small is used. Specifically, two threshold values Th1 and Th2 (Th1 <Th2) are prepared in advance. When face area size ≦ Th1, the face size = small, and Th1 <face area size ≦ Th2. Face size = medium, and if Th2 <face area size, face size = large.

  On the other hand, as the position information of the face, the ratio of the coordinate value of the center position of the face area (for example, the intersection of diagonal lines when the face area is rectangular) to the horizontal and vertical length of the image is expressed as a percentage. Use the value. In the coordinate system, the horizontal direction when the image is arranged in the horizontal direction is the x axis, and the vertical direction is the y axis. Here, assuming that the length of the image in the x direction is 100, the length in the y direction is 50, and the coordinates of the position of the center of the face area are (30, 25), the face position information is (30 , 50).

  Here, in the first embodiment, a weighting table determined in advance according to the size and position of the face is stored in the hard disk. FIG. 3 and FIG. 4 are diagrams showing weighting tables for face size and position. Note that the weighting tables LUT1 to LUT3 and LUT11 to LUT13 allow a plurality of subjects to perform a test for selecting a sample image that each subject considers preferable from a large number of sample images having various face sizes and positions. The ratio of the number of subjects who selected each sample image to the total number of subjects is used as a weighting factor. Further, in the weighting tables LUT1 to LUT3 and LUT11 to LUT13, the weighting coefficients are standardized so that the weighting coefficients for images selected by the most subjects are 1. Further, as shown in FIG. 3, the weighting tables LUT1 to LUT3 are obtained by plotting the position information in the vertical direction (y direction) on the horizontal axis and the weighting coefficient on the vertical axis for each of the large, medium, and small face sizes. is there. Further, as shown in FIG. 4, the weighting tables LUT11 to LUT13 are obtained by plotting horizontal (x-direction) position information on the horizontal axis and weighting factors on the vertical axis for each of large, medium, and small face sizes. is there.

  Then, the evaluation value calculation unit 32 selects a weighting table corresponding to the face size from the weighting tables LUT1 to LUT3 and LUT11 to LUT13 (step ST6).

  FIG. 5 is a flowchart of the weighting table selection process. The evaluation value calculation unit 32 acquires the face size information acquired by the information acquisition unit 30 (step ST21), and determines the face size (step ST22). If the face size is large, the large face size weighting tables LUT1 and LUT11 are read from the hard disk 24 (step ST23), and the process ends. If the face size is medium, the medium-size weighting tables LUT2 and LUT12 are read from the hard disk 24 (step ST24), and the process ends. Further, when the face size is small, the face size small weighting tables LUT3 and LUT13 are read from the hard disk 24 (step ST25), and the process is terminated.

  Next, the evaluation value calculation unit 32 refers to the selected weighting table and acquires a weighting factor (step ST7). In the first embodiment, since two weighting tables for the x direction and the y direction are selected, two weighting factors are acquired. Then, a temporary evaluation value of the image is calculated by adding the weighting factor (step ST8), and the temporary evaluation value is added to the current evaluation value, and the addition result is used as a new evaluation value (step ST9). Then, it is determined whether or not another face area is included in the image to be processed (step ST10). If step ST10 is negative, the process is terminated. If step ST10 is affirmed, the process target is set to the next face area (step ST11), the process returns to step ST5, and the processes after step ST5 are repeated.

  Thereby, when the image includes only one face region, the new evaluation value calculated in step ST9 is calculated as the final evaluation value. When a plurality of face areas are included in the image, a final evaluation value is calculated by adding the temporary evaluation values calculated for each face area.

  As described above, according to the first embodiment, face size and position information is acquired from an image, and an evaluation value that statistically represents an image evaluation result is calculated based on the face size and position information. Therefore, it is possible to reflect the average preference of the person who views the image in the evaluation value, and thereby it is possible to evaluate the image with higher accuracy.

  In the first embodiment, the information acquisition unit 30 acquires three types of information, large, medium, and small, as the face size information. However, the information acquisition unit 30 performs processing on the number of pixels in the face area and the entire face area image. The ratio or the like may be used as it is as face size information. In this case, when the evaluation value calculation unit 32 selects the weighting table, the face size information and the threshold values Th1 and Th2 are used to determine whether the face size is large, medium, or small. In this case, the weighting table may be obtained by linear interpolation using the face size information. For example, when the face size is between the large face size and the medium face size, the large face size weighting tables LUT1, LUT11 and the medium face size weighting table LUT2, A new weighting table may be obtained by linear interpolation with the LUT 12, and a weighting coefficient may be obtained using this.

  Next, a second embodiment of the present invention will be described. FIG. 6 is a flowchart of an image evaluation process performed in the second embodiment of the present invention. Here, in the second embodiment, the evaluation value is calculated based on the face orientation information in addition to the face size and position information. When the user gives an instruction to start image evaluation from the input unit 16, the CPU 12 starts processing. First, the image is read from the hard disk 24 (step ST31), and the initial value of the evaluation value = 0 is set (step ST32). Next, the information acquisition unit 30 detects a human face area from the image (step ST33).

  Next, the information acquisition unit 30 determines whether or not a face area is included in the image (step ST34), and when step ST34 is negative, the process ends. If step ST34 is affirmed, information on the size and position of the face included in the image and information on the direction of the face are acquired from the detected face area (step ST35). Here, the face orientation information is information indicating that the face orientation included in the face area is front or side. In addition, the direction of the face may be further determined by detecting eyes from the detected face, and when there are two detected eyes, the face is front, and when it is one, the face is horizontal. In addition, a feature amount representing the face direction may be obtained from the face area, and the feature amount may be used to determine whether the face direction is front or side.

  Then, the evaluation value calculation unit 32 selects a weighting table corresponding to the face size from the weighting tables LUT1 to LUT3 and LUT11 to LUT13 (step ST36). Next, the evaluation value calculation unit 32 refers to the selected weighting table and acquires a weighting factor corresponding to the size and position of the face (step ST37). In the second embodiment as well, two weighting tables for the x direction and the y direction are selected, so that two weighting factors are acquired.

  Furthermore, the evaluation value calculation unit 32 acquires a weighting coefficient corresponding to the face orientation (step ST38). Here, the hard disk 24 stores a function for obtaining a weighting coefficient corresponding to the face orientation. This function gives a value of weighting factor = 1 when the face orientation information is front, and weighting factor = 0.5 when the face orientation is horizontal.

  Then, a weighting coefficient corresponding to the size and position of the face is added, and a temporary evaluation value of the image is calculated by multiplying this by a weighting coefficient corresponding to the orientation of the face (step ST39). The result is added to the current evaluation value, and the addition result is used as a new evaluation value (step ST40). Then, it is determined whether or not another face area is included in the processing target image (step ST41), and if step ST41 is negative, the process ends. When step ST41 is affirmed, the process target is set to the next face area (step ST42), the process returns to step ST35, and the processes after step ST35 are repeated.

  Thereby, when the image includes only one face area, the new evaluation value calculated in step ST40 is calculated as the final evaluation value. When a plurality of face areas are included in the image, a final evaluation value is calculated by adding the temporary evaluation values calculated for each face area.

  As described above, according to the second embodiment, in addition to the first embodiment, the evaluation value representing the evaluation result of the image is further calculated using the face orientation information. It is possible to reflect the average taste of the viewer, and thereby it is possible to evaluate the image with higher accuracy.

  In the second embodiment, the temporary evaluation value is calculated by multiplying the weighting factor according to the face direction by the weighting factor according to the face size and position. The provisional evaluation value may be calculated by adding the corresponding weight coefficient to the weight coefficient according to the face size and position.

  Next, a third embodiment of the present invention will be described. FIG. 7 is a flowchart of image evaluation processing performed in the third embodiment of the present invention. Here, in the third embodiment, the evaluation value is calculated based on the information on the number of faces in addition to the information on the face size and position. When the user gives an instruction to start image evaluation from the input unit 16, the CPU 12 starts processing. First, the image is read from the hard disk 24 (step ST51), and the initial value of the evaluation value = 0 is set (step ST52). Next, the information acquisition unit 30 detects a human face area from the image (step ST53).

  Next, the information acquisition unit 30 determines whether or not a face area is included in the image (step ST54), and when step ST54 is negative, the process ends. If step ST54 is positive, information on the size and position of the face included in the image and information on the number of faces are acquired from the detected face area (step ST55).

  Then, the evaluation value calculation unit 32 selects a weighting table corresponding to the face size from the weighting tables LUT1 to LUT3 and LUT11 to LUT13 (step ST56). In the third embodiment, a weighting table determined in advance according to the number of faces is also selected.

  FIG. 8 is a diagram showing a weighting table for the number of faces. The weighting table LUT4 is determined based on the result of performing a test for selecting a sample image that each subject thinks preferable from a large number of sample images having various face numbers for a plurality of subjects. The ratio of the number of subjects who selected each sample image to the total number of subjects is used as a weighting factor. In the weighting table LUT4, the weighting factor is normalized so that the weighting factor is 1 when the number of faces is one. Further, as shown in FIG. 8, the weighting table LUT4 is obtained by plotting the number of faces on the horizontal axis and the weighting coefficient on the vertical axis.

  Next, the evaluation value calculation unit 32 refers to a weighting table corresponding to the selected face size, and acquires a weighting factor corresponding to the face size and position (step ST57). Note that in the third embodiment, two weighting tables for the x direction and the y direction are selected, so that two weighting factors are acquired.

  Furthermore, the evaluation value calculation unit 32 refers to the weighting table LUT4 and obtains a weighting coefficient corresponding to the number of faces (step ST58).

  Then, a temporary evaluation value of the image is calculated by adding a weighting factor corresponding to the size and position of the face, and multiplying this by a weighting factor corresponding to the number of faces (step ST59). The addition result is added to the current evaluation value, and the addition result is set as a new evaluation value (step ST60). Then, it is determined whether or not another face area is included in the image to be processed (step ST61), and if step ST61 is negative, the process ends. If step ST61 is affirmed, the processing target is set to the next face area (step ST62), the process returns to step ST55, and the processes after step ST55 are repeated.

  Thereby, when the image includes only one face area, the new evaluation value calculated in step ST60 is calculated as the final evaluation value. In this case, since the weighting coefficient corresponding to the number of faces is 1, the calculated evaluation value is the same evaluation value as in the first and second embodiments. When a plurality of face areas are included in the image, a final evaluation value is calculated by adding the temporary evaluation values calculated for each face area.

  As described above, according to the third embodiment, in addition to the first embodiment, the evaluation value representing the evaluation result of the image is further calculated using the information on the number of faces. It is possible to reflect the average taste of the viewer, and thereby it is possible to evaluate the image with higher accuracy.

  In the third embodiment, as in the second embodiment, the face orientation information may be further used to calculate an evaluation value representing the image evaluation result. In this case, the provisional evaluation value is obtained by adding two weighting factors according to the face size and position, multiplying this by a weighting factor according to the number of faces, and further multiplying by a weighting factor according to the face orientation. It may be calculated by doing so.

  In the third embodiment, the temporary evaluation value is calculated by multiplying the weighting factor according to the number of faces by the weighting factor according to the size and position of the face. The provisional evaluation value may be calculated by adding the corresponding weight coefficient to the weight coefficient according to the face size and position.

  Next, a fourth embodiment of the present invention will be described. FIG. 9 is a flowchart of image evaluation processing performed in the fourth embodiment of the present invention. Here, in the fourth embodiment, in addition to the weighting table corresponding to the size and position of the face, a plurality of weighting tables corresponding to the size of the target face and the number of all faces included in the image are used. Are prepared, and an evaluation value is calculated using a weighting factor corresponding to the size of the target face and the number of all faces included in the image. When the user gives an instruction to start image evaluation from the input unit 16, the CPU 12 starts processing. First, the image is read from the hard disk 24 (step ST71), and the initial value of the evaluation value = 0 is set (step ST72). Next, the information acquisition unit 30 detects a human face area from the image (step ST73).

  Next, the information acquisition unit 30 determines whether or not a face area is included in the image (step ST74), and when step ST74 is negative, the process ends. If step ST74 is affirmed, information on the size and position of the face included in the image and information on the number of faces are acquired from the detected face area (step ST75).

  Then, the evaluation value calculation unit 32 selects a weighting table corresponding to the face size from the weighting tables LUT1 to LUT3 and LUT11 to LUT13 (step ST76). In the fourth embodiment, a weighting table corresponding to the face size is also selected from a plurality of weighting tables determined in advance according to the face size and number.

  FIG. 10 is a diagram showing a weight table for the size of the target face and the number of all faces included in the image (hereinafter referred to as a weighting table for the face size and number). The weighting tables LUT21 to LUT23 are a result of performing a test for selecting a sample image that each subject considers preferable from a large number of sample images having various face sizes and the number of faces. The ratio of the number of subjects who selected each sample image to the total number of subjects is used as a weighting factor. Also, in the weighting tables LUT21 to LUT23, when the face size is medium and large, the weighting factor is 1 when the number of faces is 1, and when the face size is small, the face The weighting factor is normalized so that the weighting factor when the number is one is 1.2. Further, as shown in FIG. 10, the weighting tables LUT21 to LUT23 are obtained by plotting the number of faces on the horizontal axis and the weighting coefficient on the vertical axis for each of large, medium, and small face sizes.

  Next, the evaluation value calculation unit 32 refers to a weighting table corresponding to the selected face size and acquires a weighting factor corresponding to the face size and position (step ST77). Note that in the fourth embodiment, two weighting tables for the x direction and the y direction are selected, so that two weighting factors are acquired.

  Furthermore, the evaluation value calculation unit 32 refers to the weighting table for the selected face size and number and acquires a weighting factor corresponding to the face size and number (step ST78).

  Then, a weighting coefficient corresponding to the size and position of the face is added, and a weighting coefficient corresponding to the size and number of faces is multiplied by this to calculate a temporary evaluation value of the image (step ST79). The value is added to the current evaluation value, and the addition result is used as a new evaluation value (step ST80). Then, it is determined whether or not another face area is included in the image to be processed (step ST81). If step ST81 is negative, the process is terminated. If step ST81 is affirmed, the processing target is set to the next face area (step ST82), the process returns to step ST75, and the processes after step ST75 are repeated.

  Thereby, when only one face area is included in the image, the new evaluation value calculated in step ST80 is calculated as the final evaluation value. In this case, since the weighting coefficient corresponding to the size and number of faces is 1, the calculated evaluation value is the same as that in the first to third embodiments. When a plurality of face areas are included in the image, a final evaluation value is calculated by adding the temporary evaluation values calculated for each face area.

  As described above, according to the fourth embodiment, in addition to the first embodiment, the evaluation value representing the evaluation result of the image is calculated using the information on the size and number of the faces. It is possible to reflect the average taste of the person who sees the image, and thereby the image can be evaluated with higher accuracy.

  In the fourth embodiment, as in the second embodiment, the face orientation information may be further used to calculate an evaluation value representing the image evaluation result. In this case, the provisional evaluation value is obtained by adding two weighting factors according to the size and position of the face, multiplying this by a weighting factor according to the size and number of faces, and further weighting factors according to the orientation of the face It may be calculated by multiplying.

  In the fourth embodiment, the temporary evaluation value is calculated by multiplying the weighting factor according to the size and position of the face by the weighting factor according to the size and position of the face. A temporary evaluation value may be calculated by adding a weighting factor corresponding to the size and number to a weighting factor corresponding to the size and position of the face.

  Next, a fifth embodiment of the present invention will be described. FIG. 11 is a flowchart of image evaluation processing performed in the fifth embodiment of the present invention. Here, in the fifth embodiment, the evaluation value is calculated based on the distance information from the face located at the center of the face included in the image in addition to the face size and position information. is there. When the user gives an instruction to start image evaluation from the input unit 16, the CPU 12 starts processing. First, the image is read from the hard disk 24 (step ST91), and the initial value of the evaluation value = 0 is set (step ST92). Next, the information acquisition unit 30 detects a human face area from the image (step ST93).

  Next, the information acquisition unit 30 determines whether or not a face area is included in the image (step ST94), and if step ST94 is negative, the process ends. If step ST94 is affirmed, information on the size and position of the face included in the image from the detected face area and information on the distance from the face located at the center of the face included in the detected face area are acquired ( Step ST95).

  Then, the evaluation value calculation unit 32 selects a weighting table corresponding to the face size from the weighting tables LUT1 to LUT3 and LUT11 to LUT13 (step ST96). In the fifth embodiment, a weighting table determined in advance according to the distance from the face located at the center is also selected.

  FIG. 12 is a diagram showing a weighting table for the distance from the face located at the center. The weighting table LUT 31 is based on the result of performing a test for selecting a sample image that each subject considers preferable from a large number of sample images including faces at various positions on the image for a plurality of subjects. The ratio of the number of subjects who selected each sample image to the total number of subjects is used as a weighting coefficient. Further, in the weighting table LUT31, the weighting coefficient is normalized so that the weighting coefficient for the image selected by the most subjects is 1.25.

  In the weighting table LUT31, the distance value on the horizontal axis is normalized to 1 for the face size closest to the center. This is to prevent the distance from the face located at the center from being affected by the face size when determining the weighting table. Therefore, when the size of the face located at the center is 3 cm and the distance of the target face is 3 cm, and when the size of the face located at the center is 1 cm and the distance of the target face is 1 cm, Same weighting factor.

  Next, the evaluation value calculation unit 32 refers to a weighting table corresponding to the selected face size and acquires a weighting factor corresponding to the face size and position (step ST97). In the fifth embodiment, two weighting factors are acquired because two weighting tables for the x direction and the y direction are selected.

  Further, the evaluation value calculation unit 32 refers to the weighting table LUT31 and obtains a weighting coefficient corresponding to the distance from the face located at the center (step ST98). When only one face is included, the weight coefficient is 0.2 assuming that the distance is 0.

  Then, a provisional evaluation value of the image is calculated by adding a weighting factor corresponding to the size and position of the face, and multiplying this by a weighting factor corresponding to the distance from the most central face (step ST99). The provisional evaluation value is added to the current evaluation value, and the addition result is set as a new evaluation value (step ST100). Then, it is determined whether or not another face area is included in the image to be processed (step ST101), and if step ST101 is negative, the process is terminated. If step ST101 is affirmed, the processing target is set to the next face area (step ST102), the process returns to step ST95, and the processes after step ST95 are repeated.

  Thereby, when the image includes only one face area, the new evaluation value calculated in step ST100 is calculated as the final evaluation value. In this case, since the weighting coefficient corresponding to the distance from the face located at the center is 0.2, the calculated evaluation value is obtained by multiplying the evaluation value of the first to third embodiments by 0.2. Value. When a plurality of face areas are included in the image, a final evaluation value is calculated by adding the temporary evaluation values calculated for each face area.

  As described above, according to the fifth embodiment, in addition to the first embodiment, the evaluation value representing the evaluation result of the image is calculated using the distance information from the face located at the most center. Therefore, it is possible to reflect the average preference of the viewer who sees the image in the evaluation value, and thereby the image can be evaluated with higher accuracy.

  In the fifth embodiment, as in the second embodiment, the face orientation information may be further used to calculate an evaluation value representing the image evaluation result. In this case, the provisional evaluation value is obtained by adding two weighting factors according to the size and position of the face, multiplying this by a weighting factor according to the distance, and further multiplying by a weighting factor according to the orientation of the face. It may be calculated by the following.

  In the fifth embodiment, as in the third and fourth embodiments, the evaluation value may be calculated using information on the number of faces or information on the size and number of faces. . In this case, the provisional evaluation value is added to two weighting factors according to the size and position of the face, and is added to the weighting factor according to the number of faces included in the image or the size and image of the target face. What is necessary is just to calculate by multiplying the weighting coefficient according to the number of faces to be obtained, and further multiplying by the weighting coefficient according to the distance. Even in this case, it is a matter of course that the evaluation value may be calculated using the information of the face orientation.

  In the first to fifth embodiments, the evaluation value of the image using information on the size, position, orientation, number of faces included in the image, the distance from the most central face, and a combination thereof. In addition to this, the evaluation value may be calculated using facial expression, facial brightness, information on whether or not eyes are closed, and the like.

  Further, as shown in FIG. 13, the image evaluation apparatus 1 of the present invention includes a plurality of (here, two) evaluation values for calculating the evaluation value calculated by the evaluation value calculation unit 32 and other evaluation values for the image. You may make it provide the comprehensive evaluation part 44 which calculates the evaluation value of an image comprehensively from the evaluation value which the calculation parts 40 and 42 calculated, respectively. The comprehensive evaluation unit 44 may perform comprehensive evaluation by simply adding the evaluation values calculated by the evaluation value calculating units 32, 40, and 42, and performs comprehensive evaluation by adding weights. It may be a thing.

  As described above, the apparatus 1 according to the embodiment of the present invention has been described. However, the computer functions as a unit corresponding to the information acquisition unit 30 and the evaluation value calculation unit 32 described above, and FIGS. A program for performing processing as shown in FIG. 11 is also one embodiment of the present invention. A computer-readable recording medium in which such a program is recorded is also one embodiment of the present invention. In these cases, the weighting table may be included in the program or the same recording medium, or may be provided from an external device or a separate medium.

1 is a schematic block diagram showing the configuration of an image evaluation apparatus according to an embodiment of the present invention. Flowchart of image evaluation processing performed in the first embodiment of the present invention The figure which shows the weighting table about the size and position of a face (y direction) The figure which shows the weighting table about the size and position of a face (x direction) Flow chart of weighting table selection processing Flowchart of image evaluation processing performed in the second embodiment of the present invention The flowchart of the image evaluation process performed in the 3rd Embodiment of this invention. The figure which shows the weighting table about the number of faces Flowchart of image evaluation processing performed in the fourth embodiment of the present invention. Diagram showing weighting table for face size and number Flowchart of image evaluation processing performed in the fifth embodiment of the present invention The figure which shows the weighting table about the distance from the face located in the center most Schematic block diagram showing the configuration of an image evaluation apparatus provided with a comprehensive evaluation unit

Explanation of symbols

1 Image evaluation device 2 Memory card 12 CPU
14 system memory 16 input unit 18 display unit 20 card slot 22 compression / decompression unit 24 hard disk 26 memory control unit 28 display control unit 30 information acquisition unit 32, 40, 42 evaluation value calculation unit 44 comprehensive evaluation unit

Claims (6)

Information on the size and position of the face is obtained from an image including at least one face, and when a plurality of the faces are included in the image, from the face located at the center of the image of each face included in the image Information as a distance information of each face, and when the image includes only one face, information acquisition means for acquiring information indicating that the distance is 0 as the distance information ;
When only one face is included in the image, the first weighting function that is statistically predetermined according to the size and position of various faces is referred to, and the information about the size and position of the face is used. Calculating a first weighting factor,
The weighting factor is increased by a predetermined increase rate until the distance becomes a predetermined distance from 0, which is statistically predetermined according to the distance from the most central face, and the distance is larger than the predetermined distance. A second weighting factor corresponding to the information on the distance is obtained by referring to a second weighting function in which the weighting factor is reduced by a decrease rate smaller than the absolute value of the predetermined increase rate as the value increases. Calculating the first weighting factor weighted by a second weighting factor as an evaluation value representing the evaluation result of the image;
When a plurality of the faces are included in the image, the first weighting factor corresponding to the size and position information of the face is referred to the first weighting function for each face as a temporary evaluation value Calculating, referring to the second weighting function, obtaining the second weighting coefficient according to the information of the distance, all the temporary evaluation values weighted by the second weighting coefficient, An image evaluation apparatus comprising: an evaluation value calculation means for calculating the evaluation value by adding the faces of The information acquisition means is means for further acquiring information on the number of faces,
The evaluation value calculating means refers to a third weighting function that is statistically determined in advance according to the number of various faces and obtains a third weighting factor according to the information on the number of faces.
When only one face is included in the image, the evaluation value is calculated by further weighting the first weighting factor weighted by the second weighting factor and the third weighting factor. ,
When the image includes a plurality of the faces, the temporary evaluation value weighted by the second weighting factor is further weighted by the third weighting factor and weighted by the third weighting factor. The image evaluation apparatus according to claim 1, wherein the evaluation value is a means for calculating the evaluation value by adding the provisional evaluation value for all faces . The information acquisition means is means for further acquiring information on the orientation of the face,
The evaluation value calculation means refers to a fourth weighting function that is statistically determined in advance according to the orientation of various faces and has a weighting factor that is larger as the face is facing the front. Find the fourth weighting factor according to the information,
When only one face is included in the image, the evaluation value is calculated by further weighting the first weighting factor weighted by the second weighting factor and the fourth weighting factor. ,
When the image includes a plurality of the faces, the temporary evaluation value weighted by the second weighting factor is further weighted by the fourth weighting factor and weighted by the fourth weighting factor. The image evaluation apparatus according to claim 1, wherein the evaluation value is a means for calculating the evaluation value by adding the provisional evaluation value for all faces . The information acquisition means is means for further acquiring information on the orientation of the face,
The evaluation value calculation means refers to a fourth weighting function that is statistically determined in advance according to the orientation of various faces and has a weighting factor that is larger as the face is facing the front. Find the fourth weighting factor according to the information,
When only one face is included in the image, the evaluation is performed by further weighting the first weighting factor weighted by the second and third weighting factors by the fourth weighting factor. Calculate the value,
When the image includes a plurality of the faces, the temporary evaluation value weighted by the second and third weighting factors is further weighted by the fourth weighting factor, and the fourth weighting factor 3. The image evaluation apparatus according to claim 2, wherein the evaluation value is a means for calculating the evaluation value by adding the temporary evaluation values weighted by the above-mentioned values for all the faces . Information on the size and position of the face is obtained from an image including at least one face, and when a plurality of the faces are included in the image, from the face located at the center of the image of each face included in the image Is acquired as the distance information of each face, and when the image includes only one face, information indicating that the distance is 0 is acquired as the distance information.
When only one face is included in the image, the first weighting function that is statistically predetermined according to the size and position of various faces is referred to, and the information about the size and position of the face is used. Calculating a first weighting factor,
The weighting factor is increased by a predetermined increase rate until the distance becomes a predetermined distance from 0, which is statistically predetermined according to the distance from the most central face, and the distance is larger than the predetermined distance. A second weighting factor corresponding to the information on the distance is obtained by referring to a second weighting function in which the weighting factor is reduced by a decrease rate smaller than the absolute value of the predetermined increase rate as the value increases. Calculating the first weighting factor weighted by a second weighting factor as an evaluation value representing the evaluation result of the image;
When a plurality of the faces are included in the image, the first weighting factor corresponding to the size and position information of the face is referred to the first weighting function for each face as a temporary evaluation value Calculating, referring to the second weighting function, obtaining the second weighting coefficient according to the information of the distance, all the temporary evaluation values weighted by the second weighting coefficient, An image evaluation method, wherein the evaluation value is calculated by performing addition for the faces . Information on the size and position of the face is obtained from an image including at least one face, and when a plurality of the faces are included in the image, from the face located at the center of the image of each face included in the image A distance of each face, and when the image includes only one face , a procedure for acquiring information indicating that the distance is 0 as the distance information ;
When only one face is included in the image, the first weighting function that is statistically predetermined according to the size and position of various faces is referred to, and the information about the size and position of the face is used. Calculating a first weighting factor;
The weighting factor is increased by a predetermined increase rate until the distance becomes a predetermined distance from 0, which is statistically predetermined according to the distance from the most central face, and the distance is larger than the predetermined distance. A second weighting factor corresponding to the information on the distance is obtained by referring to a second weighting function in which the weighting factor is reduced by a decrease rate smaller than the absolute value of the predetermined increase rate as the value increases. Calculating the first weighting factor weighted by a second weighting factor as an evaluation value representing the evaluation result of the image;
When a plurality of the faces are included in the image, the first weighting factor corresponding to the size and position information of the face is referred to the first weighting function for each face as a temporary evaluation value Calculating, referring to the second weighting function, obtaining the second weighting coefficient according to the information of the distance, all the temporary evaluation values weighted by the second weighting coefficient, A program for causing a computer to execute an image evaluation method, comprising: calculating the evaluation value by adding the faces of

Images